From Motor Babbling to Purposive Actions: Emerging Self-exploration in a Dynamical Systems Approach to Early Robot Development Ralf Der 1 and Georg Martius 2 1 University of Leipzig, Institute of Computer Science, POB 920, D-04009 Leipzig, Germany der@informatik.uni-leipzig.de http://robot.informatik.uni-leizpig.de 2 Bernstein Center for Computational Neuroscience, University of Goettingen, Bunsenstr. 10, D-37073 Goettingen, Germany georg.martius@ds.mpg.de Abstract. Self-organization and the phenomenon of emergence play an essential role in living systems and form a challenge to artificial life sys- tems. This is not only because systems become more lifelike, but also since self-organization may help in reducing the design efforts in creat- ing complex behavior systems. The present paper studies self-exploration based on a general approach to the self-organization of behavior, which has been developed and tested in various examples in recent years. This is a step towards autonomous early robot development. We consider agents under the close sensorimotor coupling paradigm with a certain cogni- tive ability realized by an internal forward model. Starting from tabula rasa initial conditions we overcome the bootstrapping problem and show emerging self-exploration. Apart from that, we analyze the effect of lim- ited actions, which lead to deprivation of the world model. We show that our paradigm explicitly avoids this by producing purposive actions in a natural way. Examples are given using a simulated simple wheeled robot and a spherical robot driven by shifting internal masses. 1 Introduction Adaptation and survival in uncertain and ever changing environments are one of the key challenges in natural and artificial beings. The field has seen many impacts from life sciences, one of the directions being epigenetic and develop- mental robotics [11] trying to mimic natural ontogenesis. Moreover, the role of embodiment has become an important subject in the past decade under (i ) the practical aspect of reducing computational efforts for control by exploiting the physical properties of the robot in its environment, see [12], [9], and (ii ) the more conceptual aspect that embodied sensorimotor coordination is vital for the self-structuring of the sensor space necessary for categorization and higher level cognition, see [15], [10]. S. Nolfi et al. (Eds.): SAB 2006, LNAI 4095, pp. 406–421, 2006. c  Springer-Verlag Berlin Heidelberg 2006